"P4

"P4

I'I,

0

000

z*0

OFFICE OF NAVAL RESEARCH GRANT NOO14-67-A-0387-O007

LANGUAGE, PERSONALITY, SOCIAL, AND CROU CULTURAL STUDY AND MEASUREMENT OF THE HUMAN A-R-D (MOTIVATIONAL) SYSTEM "DEPARTMENT OF

0

PSYCHOLOGY UNIVERSITY OF HAWAII HONOLULU, HAWAII 96822

Conditioned Attraction,

Similarity, and Evaluative Meaning

Richard B. Stalling

Technical Report Number 1 August 1969

PRINcA INVEMTMAT ARTHUR W. STAATS

dim

"l•,•-

enfl.-",

DDC

L

CLEARINIGHOUS(

',

Conditioned Attraction, 0

Sinilarity and Evaluative Meaningl

Richard B. Stalling2 University of Hawaii-

Technical Report Number 1 August 1969 Supported by Office of Naval Research Grant N00014-67-A-0387-0007 Principal Investigator: Arthur W. Staats This document has been approved for public release and sale; its Reproduction in whole or in part is distribution is unlimited. permitted for any purpose of the United States Government.

ABSTRACT

Recent S-R formulations have indicated that similarity between persons functions as a UCS and that interpersonal P

attraction is a classically conditioned evaluative response. The thesis of the present study is that similarity is a correlate of evaluative meaning and that the latter rather than the former is responsible for conditioning.

The Staats

conditioning procedure was used with trigrams as CS and personality-trait adjectives as UCS.

The UCS adjectives

were previously rated on evaluation and similarity scales, and these variables were held constant across levels of each other in a 2 x 2 within Ss design. thesis, it

In support of the hypo-

was found that for 89 pretested Ss evaluation and

similarity were highly correlated (.879)

and that for the 16

Ss in the conditioning procedure an evaluative response to trigrams was influe-nced by evaluation (p similarity (p > .20).

.005)

but not by

ei

In an effort to translate the concept of interpersonal attraction into S-R terminology, a number of theorists (Byrne, in press; Lott & Lott$ 1960; Staats, 1964) have proposed that attraction be consiurl&.

one of a class of implicit attitude

responses (.icob, 19,17).

Byrne (in press) and Staats(1958,1968)

have further specified classical conditioning as a model of the process by which these implicit responses are formed.

In

general, the classical conditioning models suggest that a person originally functions as a relatively "neutral" stimulus which, when paired with pleasant or unpleasant events as

UCS, comes to elicit an implicit evaluative response or CR. once formed, the stimulus components of these implicit CR's are thought to mediate overt behavioral responses employed as dependent measures Of attraction, such as approachavoidance, sociometric and other ratings, and verbal assessments. To incorporate within an S-h framework findings indicat-

ing that similarity between people results in attraction (e.g., Backiman & Secord, 1964; Newcomb, 1956,

1961).

Clore

and Byrne (in press) have proposed a separate category of the conditioning moco as a UCS.

1

in which similarity is assumed to function

Using a procedure closely analogous to classical

aviacscoercadoheIaigadvra ses conditioning,

these researchers demonstrated that photographs

of unknown persons, following pairing with attitude statements either like or edlike those of the subject, evoked positive and negative affective revlaiponses.

rshile real

individuals were not used as stimuius Luects, this study

2 suggests,

as do studies using a cognitive-ba!ance model,

atr action occurs between persons holding It

-

that

.r attitudes.

also suggests the possibility that the formation of attrac-

tion responses involves a process of classical conditioning C

in which similarity acts as UCS.

In a series of other studies

employing an S-R interpretation, Byrne and his associates have also found attraction to be a function of similarity with regard to such variables as economic status (Byrne, Clcre,

& 10orchel,

SStefaniak,

1966),

1967),

personality traits (Byrne,

Griffitt,

and self-concept (Griffitt, 1966).

While the similarity dimension has accurately predicted the formation of an affective response,

it

seems reasonable

to propose that most individuals consider most of the attitude statements and many of the personality trait descriptions which characterize them to be good or pleasant and the ones which dr

1 iot

to be bad or unpleasant.

For those verbal

stimuli in which this is the case one could as reasonably ascribe the positive UCS value of these stimuli to pleasantness as to similarity.

In other words,

the predictive value

of the similarity dimension may be attributable to a correlation between sinilarity and a measure such as the evaluative dimension (e.g., pleasant-unpleasant,

good-bad) of the

semantic differential. As part of a general learning theory of human behavior, Staats (see Staats,

1966; Staats & Staats,

1963) has suggested

that evaluative rneaning is the appropr'ate measure of stimuli used to condition an evaluative

(attraction) response

3 to a person or other discriminable stimulus. dimension is at least a sufficient,

That such a

if not exclusive, pre-

dictor of affective conditioning has been shown in a series of studies (e.g., 4 Heard,

1959).

Staats $ Staats, 1957, 1953; Staats, Staats, Based on Staats'

view of attitude condition-

ing, the thesis of the present ztudy is that similarity does not comprise a separate category of reinforcement but derives its apparent UCS value from the fact that it correlate of evaluative meaning.

is a

The strategy of the study

was to examine the power of personality-trait adjectives to condition evaluative meaning to trigrams.

It was expected

that when evaluation and similarity were held constant across levels of each other, evaluation would be the sole predictor of UCS value. Method Subjects Using the criteria described below, twenty-four students attending the University of HaAqii were selected to partictpate in the conditioning procedure,

eight of whon were

eliminated following a post-experiziental assessment of awareness. educaxtoio

Of these 16 subjects 13 were enrolled in graduate courses and consisted primarily of grade school

and high school teachers. males.

Twelve of these thirteen were fe-

The remaining subjects were undergraduate students

in introductory psychology courses, All subjects were volunteers,

two hales and one female.

*mut tie undergraduates

4 received points contributing to the grade in the course from which they were solicited. The pretest booklet The pxetesW

booklet consisted of 121 adjectives selected

from Anderson's (196G)

of 555 personality-trait adjec-

list

W1ords were randomly assigned to 11 page!,

tives.

same randomization was used for all subjects. word were two sever-point scales,

and the

Beneath each

the evaluation and simi-

Th• two scales always appeared in the same

larity scales.

order, evaluation followed by similarity,

as shown in this

example:

AGGRESSIVE pleasant like me

-

:

:

:

:

:

:

:

-: :

:

unpleasant

:

unlike me

General instructions on the cover sheet of the booklet were similar to those used by Osgood (in baum,

Osgocd, Suci, & Tannen-

1957).

Selection of UCS words and subjects Based on the ratings of words on the two scales, lists of 12 adjectives per list

four

were constructed for each

subject which had the following characteristics:

List 1 wzs

comprised of words rated both pleasant and like me, List 2 of words rated unpleasant but like me,

List 3 pleasant but

unlik• me, and List 4 unpleasant and unlike me. words to the tou. lists, ratings of 1, 2,

In assigning

or 3 on the

5 seven-point (1 to 7) scales were considered pleasant or like me and ratings of 5, unlike me.

6,

or 7 were considered unpleasant or

Occasionally ratings of 4 (neutral) were used on

one of the two dimensions to codiplete the list of 12 adjectives.

Only those subjects whose protocols provided the requisite four lists were selected to participate in the conditioning procedure. just described.

Of 103 pretested subjects,

38 fit

the criteria

As previously mentioned, not all 38 were

employed as subjects in the conditioning procedure. Figure 1 summarizes the description of the lists, illustrates the 2 x 2 within subjects design,

and shows the mean

ratings of the UCS adjective lists for all subjects on the evaluative and similarity dimensions.

It was desired that

each of the two rated dimensions be held constant across each of the two levels of the other factor.

Reading to the right

of the diagonals and down shows the mean values of pleasant and unpleasant words to be quite close across levels of similarity, and reading to the left of the diagonals and across shows the mean value!, of the like me and unlike me words to be quite close across levels of evaluation. Procedure Subjects participated in the conditioning phase one at a time betwpe.

five and ten days after the pretest.

trigrams (YOF,

LAJ,

.J, and XFi)

Four

functioned as CS syllables

and were presented by means of a slide :rojector set so that each slide would automatically be shown for five seconds with

Fig. 1.

Mean ratings of the UCS adjective lists on the evaluation and

similarity dimenasions.

(Evaluative ratings are to the right and similarity

ratings to the left of diagonal.)

EVALUATIOtN Pleasant 1.64 :2:.2 8

Like Me

Unpleasant 6.43 2.2

Ulike Me

List 4

List 3

-- -'

t-

--

-no~•

"' --

1.66|6.47

-.

6 a five second interval between slides.

Approximately one

second after the onset of each trigram the UCS words were spoken by the experimenter and then repeated aloud by the subject, All 12 adjectives comprising a particular list were consistently paired with only one of the four trigrams.

For

example, for a particular subject YOF might be paired with pleasant and like me adjectives, LAJ with pleasant and unlike me adjectives, etc.

Each of the 12 adjectives was

presented only once for each subject, and each trigram was thus presented 12 times for a total of 48 trigram-adjective pairings.

To counterbalance possible differences between

trigrams each trigram was paired equally often with each of the four lists of adjectives across the 16 subjects. The order in wlich the trigram adjective pairs were presented to the subjects was counterbalanced by using Latin squares.

The 48 slildes were ordered so that each of the

trigrams occurred once in each block of four slides and no

trigram occurred more than once in a row. Instructions to subjects indicated that the study was a learning task in which different modes of presentation, auditory and visual, would be examined.

Following presen-

tation of the 48 slides each subject was asked to rate the pleasantness-unplea3antneýs

of the four trigrams under the

guise that the way he felt about these ýyllables might influence learning.

The four trigrants and seven-point

evaluative scales were presented in random order for each

7 subject on separate pages.

Recall of the words was then

tested and the subjects were asked to write on the back of their booklet any thouqhts which occurred to them in the course of the experiment which might be relevant, particularly regarding the purpose of the experiment. Eight of the subjects indicated that pleasant or unpleasant words (or words of a particular feeling) had been associated with particular trigrams.

To preserve the counter-

balanced design, each time a subject was judged aware additional subjects were run until the blocks were complete. None of the subjects indicated awareness of the similarity dimension, that trigrams had also been paired with words which were like them or unlike them. Results Table 1 presents the mean evaluative ratings of the CS trigrams following conditioning and the analysis of variance. Summing over levels of the two variables produced mean values which are quite close to a "neutral" rating of four for the similarity factor and in the expected directions above and below four for the evaluative factor.

In the analysis of

variance the F for evaluation is highly significant (p - .005)

while that for similarity is not (p > .20).

The

similarity by evaluation interaction i.,also not significant but a tendency toward interc:t1ion is r-:.dent (p < .10). The mean ratings of the CS trivr-.',:

for subjects itdged

aware, shown in Table 2, reflect the same relationships as

8 do the data for unaware subjects with the difference between Since

levels of the evaluative dimension greatly increased.

counterbalancing of trigrams for aware subjects could not be accomplished no formal statistical analysis was carried out, but occurrence of the same trend toward interaction in both aware and unaware subjects urges caution in rejecting such a factor. The mean ratings by 89 subjects3 on the two scales for each of the 121 words was determined and a single correlation between the maean evaluation and similarity ratings of these

words was found to be

879. Discussion

Byrne and his associates (Byrne, Byrne,

in press; Clore

in press) have suggested that similarity between per-

sons with regard to such variables as attitudes and personality traits functions as a UCS in conditioning attraction. On the basis of the results of the present study, it

is

suggested that evaluative meaning is a more direct and parsimonious measure than is similarity of the power of stimuli to condition an affective response.

As hypothesized,

when evaluation and similarity were held constant across levels of each other, evaluation accurately predicted conditioning of attitudes to trigrams, contribute to this prediction. tion (.879)

and similarity did not

The sizable positive correla-

found between these dimensions may account for

the apparent UCS value of similarity in some of the previous

9 studies of conditioned attraction. while the experimental situation was, compared with typical attraction studies, highly artificial, the results suggest a clarification of the classical conditioning model which is in some respects quite congruent with cognitive and other theories of attraction. using a balance model (1953) terminology (1956),

For example, Newcomb,

first

and later using reinforcement

has also maintained that similarity of

attitudes produces attraction; but he has specified that it has this effect only because the likelihood of reward is higher for similar than dissimilar persons.

Thibaut and

Kelley (1959) have also suggested that it is the reward available in a relationship which results in attraction and that in some cases reward depends on sinilarity and in others on differences between persons.

If,

as seems reason-

able, the reward to which these theorists refer elicits positive evaluative meaning, similar relationship.

the present findings reflect a

W-hile a correlation between evalua-

tive and similarity was found, it appeared to be the evaluative rather than the similarity component which conditioned an affective response. Byrne and his associates (Byrne,

Griffitt, 4 Clore, 1968)

have also related evaluative meaning to the similarityattraction relationship by specifying that similar attitude statements have "evaluative (affective) meaning and hence reinforcement properties (p.

962)."

One implication of this

statement is that similar and dissimilar attitude statements

10 always or almost always connote positive and negative evaluative meaning, the case.

and it

seems reasonable to suggest that this is

tJith regard only to attitudes,

it

may be that in

the process of agreeing or disagreefng with attitude statements subjects are also categorizing them as positively or negatively evaluative.

For example,

if

subjects agree that

"Medicine should be socialized" they are probably also indicating that this is

a good (positively evaluative) atti-

With regard to personality traits, however,

tude.

the r::.

present study (as well as previous research on self-ideal discrepancy) clearly indicates that subjects frequently dislike the traits which happen to characterize them and like traits which do not.

Despite the high overall correlation

between the means of the similarity and evaluation measures, there was enough individual variation so that 37% of the original sample (38 of 103 subjects) rated at least 20% (24 of 121) of the adjectives listed as being either like me but unpleasant or unlike me but pleasant. for similarity of personality traits, it to suggest,

like Thibaut and Kelley,

Thus,

at least

seems appropriate

that reward (and hence

attraction) may in some cases depend on similarities and in others on differences between people. It

is

important to note that the concept of attraction

in the present study was specifically restricted to a definition in terms of an evaluative rating.

It seems likely

that other measures of attraction, such as frequency of interaction, may be a function of personality similarities

For example,

which are independent of evaluative meaning.

two persons who possess the same low-valued personality traits might choose each other in a social or work situation. since they night have learned to expect more reward and less censure from similar persons than from more attractive perThe distinction being drawn h!.re is between attraction

sons.

considered as an evaluative response,

as it

is

in the

present paper, versus attraction as expected reciprocation in friendship.

This distinction is similar tG the one

Newcomb (19S6)

makes between "admiration at a distance'

anA

frequency of interaction in friendship groups. The orientation of the present paper is not intended to serve simply as a translation of terms from one paradigm to another but to represent an approach to attraction which emphasizes language and communication between persons.

It

is suggested that the mere quantity of affective responses to words occurring during inter-personal interaction combines (not necessarily in an additive manner) to produce responses of attraction or dislike.

It would be predicted,

for

example, that with such variables as word arrangement (e.g., see iieise, 1969) and non-verbal cues held constant,

the

mere contiguous occurrence of positive or negative affective words with a person would account for a large portion of affective responses to the person.

The implications of this

position are presently being studied by the author and A. Staats in a situation involving an actual individual as CS object.

12 The correlation between evaluation and sirilarity found in the present study indicates that many people tend to view the personality traits they possess as good traits.

The

development of this aspect of human personality could be accounted for by assuming that parents reinforce those responses of the child which are most reinforcing to the parents and that, in general, parents prefer and reinforce the child's pleasant rather than unpleasant self-descriptions.

In addi-

tion, Staats (1968) would suggest that in the process of instrumentally conditioning verbal descriptions one is also classically conditioning Peaning to the words used,

and in

so doing establishing the capacity of these verbal stimuli to reinforce an instrumental response. Both Byrne and Staats have found that words or phrases capable of classically conditioning an evaluative response are also able to function as reinforcement and punishment in an instrumental task (Golightly & Byrne, 1964; Finley & Staats,

1967).

theoretical orientations,

1964; Staats,

As might be expected from their

Staats employed words of positive

and negative evaluative meaning as reinforcers and aversive stimuli, while Byrne employed similar and dissimilar attitude statements.

It may, as discussed above, be difficult

to separate the evaluative or similarity components of attitude statements.

On the basis of the results of the

present study, however,

it

would be predicted that similarity

of personality-trait words would not function as a reinforcer in an instrumental task beyond the extent of their correlation with evaluative meaning.

13 References Anderson, N. H.

Likeableness ratings of 555 personality-

trait adjectives.

Journal of Personality and Social

Psychology, 1968, 9, 272-279. Backman, C. 11.,

G Secord, P. F.

The compromise process and

the affect structure of groups. 17

Human Relations, 1964,

19-22.

Byrne, D.

Attitudes and attraction.

In L. Berkowitz (Ed.),

Advances in Experimental Social Psychology. New York: Byrne, D.,

Academic Press,

Clore, G. L.,

Vol. 4,

in press.

Jr., & tterchel, P.

The effect of

economic similarity-dissimilarity on interpersonal attraction.

Journal of Personality and Social

Psychology, 1966, 4, 220-224. Byrne, D.,

Griffitt, 11.,

& Clore, G. L., Jr.

Attitudinal

reinforcement effects as a function of stimulus homogeneity-heterogeneity. Verbal Behavior, 1968, Byrne, D.,

Griffitt, '. B.,

Journal of Verbal Learning and 7, 962-964. & Stefaniak, D.

Attraction and

similarity of personality characteristics.

Journal

of Personality and Social Psychology, 1967,

5, 82-90.

Clore, G. L., Jr.,

4 Byrne, D.

interaction.

In R. B. Cattell (Ed.),

personality study. Doob, L. 11.

The process of personality

Chicago:

Aldine, in press.

The behavior of attitudes.

view, 1947, 54,

135S-156.

Handbook of modern

Psychological Re-

14 Finley, J.

& Staats, A. 11.

R.,

Journal of Verbal Learning and

reinforcement stimuli. Verbal Behavior,

1967,

6,

& Byrne, D.

Gollghtly, C. C.,

Evaluative meaning words as

193-197. Attitude statements as positive Science,

and negative reinforcements. Griffitt, 11. B.

1964,

146,

798-799.

Interpersonal attraction as a function of

self-concept and personality similarity-dissimilarity. Journal of Personality and Social Psychology,

1966,

4,

S81-S84. Affectual dynamics in simple sentences.

Heise, D. R.

of Personality and Social Psychology, Lott, B. E., & Lott, A. J.

Newcomb,

T. H.

204-213.

3j

The formation of positive

attutides toward group members. Social Psychology,

1969,

Journal

1966,

61,

Journal of Abnormal and

297-300.

An approach to the study of communicative acts.

PsychologicalReview, 1953, 60, 7'93-404. Newcomb,

T. M.

The prediction of interpersonal attraction.

American Psychologist, Newcomb,

1956,

11, 57: S86.

T. IU. The acquaintance process.

NJew York:

Holt,

Rinehart & :iinston, 1961. Osgood,

C. E.,

Suci, G. J.,

went of meaning. Press,

& Tannenbaum,

Urbana,

Ill.:

P.

I,. The measure-

University of Illinois

1957.

Staats, A. r1. Winston,

HIuman learning.

New York:

Holt,

Rinehart 4

1964.

Staats, A. '.. Learning language, hart & Winston,

196S.

and cognition.

Holt, Rine-

15 Staats, A.

1.

• Staats, C. K.

classical conditioning.

Attitudes established by Journal of Abnormal and

Social Psychology, 1958, s7, 37-40. Staats, A. I&., New York:

Staats, C. X. Cqmplex human behavior. S Holt, Rinehart, $ W/inston,

Staats, A. '0., Staats, C. K.,

4 heard,

1963.

:. G.

Language

conditioning of meaning to meaning using a semantic generalization paradigm. Psychology, 1959, Staats, C. K.,

57,

Journal of Experimental

187-192.

4 Staats, A. V?.

classical conditioning.

Meaning e!tablished by Journal of Experimental Psy-

chology, 1957, 54, 74-79. Thibeut, J. 11., 4 4elley, H. H. The social pyvchology of groups.

K

New York:

Wiley, 1959.

__

Table 1 Mean Evaluative Ratings of CS Trigrams per Condition:

Similarity

Unaware Ssa

Evaluation Total

Pleasant

Unpleasant

Like me

3.56

4.06

3.81

Unlike fe

2.88

5.00

3.94

Total

3.22

4.53

3.88

Analysis of Variance Source

df

Evaluation (A)

1

Similarity (B)

1

Subjects (S)

_iS

31.64

15 1

8.23

AXS

15

2.31

BXS

iS

4.54

AXB X S

15

2.40

Total

63

al is pleasant end of scale, 7 is unpleasant. < .005

13.70*

.05

AX B

*.

F

3.43

Table 2 Mean Lvaluative Ratings of 'S Trigrams per Condition:

Aware Ss

Evaluation

Similarity

Total Pleasant

Unpleasant

Like me

2.00

5.38

3.69

Unlike me

1.38

6.38

3.94

Total

3.22

4.53

3.38

Footnotes iThis article is based on the author's doctoral dissertation submitted to the Department of Psychology of the University of Hawaii.

The author wishes to express his appreciation

to A. Staats, committee chairman, for his valuable advice and guidanre; to the other members of his committee, with a particular note of thanks to K. Viake and I. Reid; to Wi. Cash and I. Reid for their help in finding volunteer subjects from education; and to D. Byrne and 1W.Griffitt for their very helpful comments on the proposed as well as the completed project. 2 Now at Bradley University. 3 Fourteen

subjects who did not complete the pretest

booklet were eliminated from the analysis.

DOCUMENT CONTiROL DATA. (eS

llas

urity

ORIGINATIN'G

ACTIVITY

t!or

of title, h,,dy, of obxract asid indexietuf annotation muot he enfelred when the uverall :epor.t is c lassified) 20. REPORT

(Crp.orate lowthur)

Dr. Arthur W. Staats Department of Psychology, Honolulu, Hawaii 96822 -1 REPORT

?T

R&D

SECCIRITY

CLA•$S1'ICAVION

Unclassified OUP

University of Hawai

TLE"

Similarity and Evaluative Meaning

Conditioned Attraction, NOTES (Type oft

4

DESCRIPTIVE

.I

Ao, VHORi$I (First

eport and inclusive dotes)

name, meddle Inltiat, 1eet name)

Richard B. Stalling 6

Te. TOTAL

IFFPIRT DATE

August 190 oeCONTRACT OR GRANT 14O ONR N00014-b7-A-0387-0007 . PROJCT

NO,

OF PAGES

-

7

. NO

Sa. OReIGI;JATOR*S

OF REr•

23

2O REPOAT NUJMB

tRIS)

Technical Report Number 1

NO

Ob. OTHER REPORT NOMSt (Any other report) this

C.

niambere thet may be *eul9n•d

d. t0

DISTRIBUTION

$'ATEMEP-NT

Distribution of this document is unlimited. II

12

TLJPPLEMENTARY NOTES

SPONSORING

tILl I

A

Y k CI 4•tITY

Office of Naval Research Washington, D. C. 20360 3

A•STRACT 1

Recent S-R formulations have indicated that similarity between persons functions as a UCS and that interpersonal attraction is a classically The thesis of the present study is conditioned evalutive response. that similarity is a correlate of evaluative meaning and that the latThe Staats ter rather zhan the former is responsible for conditioning. conQitioning procedure was used with trigrams as CS and personalityThe UCS adjectives were previously rated on trait adjectives as UCS. evaluation and similarity scales, and these variables were held conIn stant across levels of each other is a 2 x 2 within Ss design. support of the hypothesis, it was found that for 89 jretested Ss evaluation and similarity was highly correlated (.879) and that for the 16 Ss in the conditioning procedure an evaluative response to trigrams was influenced by evaluiion (p < .OOS) but not by similarity (p , .20).

DDFORM 1473 11o 7 DD,

I

•

•,

G

l.

!.•0•

,•'

!

,,'

t~r!•

I•.•|l

,•

,

t

14

LINK NOLEC

£

LINK WT

ROLE

V, WT

F|,

attitudes classical conditioning

evaluative meaning human learning and conditioning interpersonal attraction social interaction social learning

DD (PAGE

F01..1 2)

473

B3ACK)_________ S

'CWityCisalfIfcetion

LINK ROLE

C WT